Corrective Feedback in EFL university classrooms: A Case Study at an Italian … · 2017-12-14 ·...

Questo articolo esamina gli errori commessi dagli studenti e le correzioni fornitedurante le ‘esercitazioni orali’ in un corso di inglese svolto in un’università italiana.Il corpus è costituito dalle trascrizioni di sei lezioni offerte a due gruppi, uno prin-cipiante e l’altro pre-intermedio. I risultati evidenziano che sebbene i ‘recast’ sianoil tipo di correzione più usato in entrambi i gruppi, la distribuzione degli altri tipi dicorrezione varia. Anche la frequenza delle correzioni varia tra i due gruppi e risulta,infatti, molto più alta nel gruppo di principianti. Inoltre, l’obiettivo pedagogico e lediverse modalità di partecipazione degli studenti alle attività hanno entrambi in-fluenzato la percentuale degli errori corretti e il tipo di correzione fornito. Non sor-prende che gli errori commessi durante le attività grammaticali siano stati corretticon maggiore frequenza rispetto a quelli commessi durante le attività comunicative,con una differenza particolarmente significativa nel gruppo pre-intermedio. Infine,lo studio evidenzia il ruolo degli studenti nel fornire correzioni e suggerisce chequesto aspetto possa essere oggetto di un’ulteriore ricerca.

1. Corrective Feedback

Although considerable research has been devoted to correctivefeedback given to students learning another language, there is still noclear answer as to how teachers should deal with the errors studentsmake during oral classroom work. Early studies on this issue illustratedits complexity and brought attention to the inconsistency andimprecision shown by teachers (Allwright 1975, Chaudron 1977,Fanselow 1977, Long 1977). One of the problems is that students often

45

JEAN JIMENEZ 1

Corrective Feedback in EFL university classrooms:A Case Study at an Italian university

1 I wish to thank Kris Spelman Miller, School of Linguistics and Applied Language Studies,The University of Reading, for her guidance and support throughout this project, which fulfilledpart of the requirements for the degree of MA in TEFL. I would also like to thank my colleagueand friend, Susie Caruso, for allowing me to observe her lessons and taking the time to answer allof my questions.

do not realize that they are being corrected. Therefore, the first step is forlearners to become aware that they have made an error. However, asBrown (1987) underlines, corrective feedback must not consist inconstant interruptions or in punitive reinforcement as this would lead tostudents giving up attempts at communication (Vigil and Oller 1976).Recent studies have suggested the use of feedback techniques which maybe more easily noticed by learners and which may be more effective(Lyster and Ranta 1997, Lyster 1998, Panova and Lyster 2002).

In their study on corrective feedback and learner uptake in Frenchimmersion classes in elementary schools, Lyster and Ranta (1997)identified six different feedback types: recasts, explicit correction,elicitation, metalinguistic clues, clarification requests and repetition.2Although recasts were the most widely used technique, making up 55%of the feedback moves, they had the lowest rate of uptake (31%). Lysterand Ranta attributed this low rate of uptake to the fact that students maynot always recognize recasts as corrections. In fact Lyster (1998: 206)explains that recasts “risk being perceived by learners as alternative yetequally correct forms”. Similar results were found in a study on adultlearners’ perceptions of interactional feedback carried out by Mackey etal. (2000), who point out that recasts were often perceived as alternativeways of saying the same thing. In fact, they highlight that recasts wereused for morphosyntactic errors, but students seldom recognized thetarget of the feedback. On the other hand, phonological and lexicalerrors were followed by negotiation or negotiation and recasts, andfeedback was perceived more accurately. Because of the ambiguity ofrecasts, Lyster and Ranta (1997) suggest that teachers use moreelicitation, metalinguistic clues, clarification requests and repetition,since these techniques are not only more likely to be perceived ascorrective feedback, but, unlike both recasts and explicit correction,they encourage student generated repair. In fact, they point out thatwhereas both recasts and explicit correction provide the correct form,the other four techniques can be classified as ‘negotiation of form’, thatis “the provision of corrective feedback that encourages self-repairinvolving accuracy and precision and not merely comprehensibility”

46

Linguistica e Filologia 23 (2006)

2 These will be examined in more detail in the next section, where examples of each will beprovided.

(Lyster and Ranta 1997: 42). However, there are also problems withrepetitions. Chaudron (1998) explains that because teachers often alsorepeat students’ correct utterances to show agreement, appreciation orunderstanding, repetitions of incorrect forms may not always beperceived as corrective feedback. He therefore suggests that teachersshould repeat incorrect utterances with more emphasis or with adifferent intonation in order to make them more explicit.

Turning back to recasts, some studies have shown that recasts canbe effective as long as certain conditions are met. These are “(a)individualized attention, (b) consistent focus, (c) developmentalreadiness, and (d) intensity” (Han 2002: 568). Furthermore, the results ofa study on the effects of recasts on the production of question forms inEnglish carried out by Mackey and Philp (1998) also suggest thatdevelopmental readiness may play a role in the effectiveness of this typeof feedback. In fact, they found that for “more advanced learners,interaction with intensive recasts may be more beneficial than interactionalone in facilitating an increase in production of targeted high levelmorphosyntactic forms” (1998: 339). Moreover, in a study on errortreatment in adult ESL classrooms, Panova and Lyster (2002) found thatrecasts accounted for the highest number of feedback moves (55%), butstudents’ uptake after recasts was only 40%. Nevertheless, they arguethat recasts are important in classroom discourse since they are “efficientand natural ways of responding to students” as well as a good way ofdealing with incorrect utterances which “are beyond the students’ currentinterlanguage” (2002: 591). In fact, Lyster and Ranta (1997) found that,although recasts were the most popular type of feedback regardless ofproficiency level, the teacher with the most advanced students usedrecasts considerably less. This implies that teachers may find recastsmore useful with students of lower proficiency level who may not yet beready to produce the correct target form.

Other researchers underline the importance of trying to matchcorrective feedback to individual students’ preferences and as aconsequence suggest the use of a wide range of feedback techniques(Allwright 1975, Tedick and de Gortari 1998, James 1998, Allwrightand Bailey 2004). However, it is not always easy to satisfy students,who may prefer explicit corrections (Kaufmann 1993, cited in James1998), though these are not as effective. In fact, Bartram and Walton

47

J. Jimenez, Corrective Feedback in EFL university classrooms: A Case Study at an Italian university

(2001) believe that the least effective type of corrective feedback issimply providing students with the correct form.

Researchers have also stressed the importance that self-correction aswell as peer-correction can have in language learning (Chaudron 1988,Van Lier 1988, Slimani 1992 (cited in Panova and Lyster 2002), Tedickand de Gortari 1998, Bartram and Walton 2001) First of all, learnersmay benefit more from the errors they correct themselves. In fact, in astudy by Slimani (1992, cited in Panova and Lyster 2002: 576), learnerssaid that they noticed “forms that they were pushed to self repair morethan forms that were implicitly provided by teachers”. Lyster and Ranta(1997: 57) believe that student-generated repair is important because itgives learners opportunities “to automatize the retrieval of targetlanguage knowledge that already exists in some form” and rely on“their own resources and thus confront errors in ways that may lead torevisions of their hypothesis about the target language”. Thus, bylearning to recognize and correct their own errors, students becomemore independent. Therefore it is very important that teachers givestudents enough time to self correct instead of immediately interruptingthem with the correct answer. James (1998) also suggests that, sincecorrection should be aimed at the whole class,3 teachers should involveother students in the correction of each other’s errors as it might be away of taking away some of the embarrassment of being explicitlycorrected by the teacher. Finally, regardless of who provides correctivefeedback, an atmosphere of mutual support and cooperation can makethings easier, more enjoyable and more fruitful.

The pedagogical focus of the lesson is also an important factor indeciding when an error should be corrected. Studies reviewed byChaudron (1998) show that when instructional focus is on form,corrections occur more frequently. Moreover, he explains that teacherswill be more likely to correct an error when it pertains directly to thefocus of the lesson. Not surprisingly, a high priority is given to errorcorrection in grammar-based instruction whereas error correction doesnot constitute a major proportion of the activity in L2 classrooms whichfocus on communicative activities. This reflects Hendrickson’s

48


3 James refers to EFL classes in general and gives no specifications regarding class size or age.

(1978:157) suggestion that teachers should be more tolerant of errorsmade during communicative practice and “should reserve errorcorrection for manipulative grammar practice”, a view which mostpractitioners still agree with.

Another important factor to consider when deciding whether or notcorrective feedback should be given is the learner’s readiness toincorporate the feedback provided (Allwright and Bailey 2004, Klein1986 (cited in Tonkyn 1998), Tonkyn 1998). The learners’ proficiencylevel should also be taken into consideration when deciding what kindof feedback to provide (Lyster and Ranta 1997, Tedick and de Gortari1998).

Teachers must also decide which errors to correct. Hendrickson(1978) recommends focusing more on errors that impede intelligibility(global errors), and not worrying too much about errors which do notaffect the meaning of the message (local errors). He also advisesteachers to correct errors which have a stigmatizing effect as well aserrors that are frequent. However, research shows that thesesuggestions, especially the last one, are not always followed. Forexample, through a review of studies carried out in L2 classrooms,Chaudron (1998) found that the more a particular type of error is made,the less likely a teacher is to provide corrective feedback. In fact hefound that discourse, content and lexical errors received more correctivefeedback than grammatical and phonological errors, which were muchmore common. He also found a wide variation between the overallpercentage of errors treated by different teachers. Similarly, in a studycarried out by Lyster (1998) on the relationship between error types,feedback types and immediate learner repair in four Canadian Frenchimmersion classrooms, results show that teachers provided morecorrective feedback after phonological and lexical errors, 80% and 70%correction rate respectively, than after grammatical errors andunsolicited uses of L1, 56% and 46% respectively. Therefore, althoughgrammatical errors made up the highest percentage of errors (50%),they did not receive the highest rate of feedback. Moreover,phonological errors only made up 16% of the total errors and yetreceived a high rate of feedback by all four teachers.

49


2. Research methodology

2.1. Research aims

As outlined above, many studies have been devoted to correctivefeedback. Some studies have concentrated on corrective feedback andlearner uptake in content based instruction in French immersion classes(Lyster and Ranta 1997, Lyster 1998), in ESL adult courses (Panova andLyster 2002) and in adult ESL and Italian as a foreign language courses(Mackey et al. 2000). Other studies have focused more specifically onrecasts in either middle school ESL science courses (Doughty andVarela 1998) or ESL adult courses (Han 2002, Mackey and Philp 1998).Finally, the differences between negative feedback provided to adultand children ESL learners have also been investigated (Oliver 2000).However, I thought it would be of interest to look into how correctivefeedback is provided in EFL contexts, in particular at university level.Thus, this study focuses on the types of errors made and on thecorrective feedback provided during oral classroom work in an Italianuniversity. It seeks to investigate whether proficiency level andpedagogical focus can have an effect on the type of corrective feedbackprovided and on the types of errors treated. It also examines the rolestudents themselves play in providing feedback.

In particular, the study aims to answer the following questions:

1. When listening to students’ utterances, do teachers correct differenttypes of errors depending on what kind of classroom activity is beingcarried out?

2. How are these errors corrected?3. Does error treatment also depend on the students’ proficiency level?

2.2. Teaching context

The teaching context selected as the site for data collection is an EFLcourse held at the Faculty of Economics, University of Calabria (Italy).The course is held over two 5-week-periods, with a 3-week-exam periodin between, with one-hour lessons held three times a week for a total of30 hours in class. It is structured around an integrated syllabus whichemphasizes the importance of structures and how they are related to their

50


communicative function. Throughout the course, emphasis is given tospeaking, listening and reading skills, with little work on writing skills.Students are divided into classes depending on their level of English,which is established through a written placement test and usually variesfrom beginner to pre-intermediate. However, the course syllabus is thesame for every class since all students are required to take the same finalexam regardless of the class they belong to.

In order to guarantee a clear difference in the proficiency level of thestudents involved, an important criterion as regards my third question,the classes observed were an elementary level class (from now onreferred to as Group A) and a pre-intermediate level (Group B). Bothclasses were held by the same teacher.

2.3. Participants

Group A was made up of 38 native Italian speakers, all of whom hadstudied English before but had a very limited knowledge, as wasevidenced in the placement test. They were therefore grouped into anelementary level class. Group B consisted of 35 native Italian speakersas well as two foreign students, one from Poland and one from Albania.They had achieved the highest scores on the placement test (pre-intermediate level). Not all of the students attended every lessonobserved, with the average number of students present during thelessons observed being about 28.4 Attendance was slightly lower forboth groups during the last lesson observed, probably due to the factthat it was the last week of lessons when attendance rates are usuallylower, and this might have influenced classroom behavior.

The teacher is a 30 year-old female native speaker of English with 8years’ teaching experience. She is a very enthusiastic, outgoing andfriendly teacher and was selected for three main reasons. First of all, shewas willing to have her lessons observed and video-taped. Secondly, itwas important to observe the same instructor teaching the same courseto groups of different proficiency levels (question 3) since involvinganother teacher would have meant carefully having to take intoconsideration factors such as school of thought, skill, experience,

51


4 Attendance Group A- lesson 1: 31, lesson 2: 30, lesson 3: 25; Group B- lesson 1: 31, lesson2: 29, lesson 3: 23.

personality, all of which may affect which errors are corrected and howthey are corrected. Finally, it was also important to observe groups thathad the same number of students since teachers and students may actdifferently in larger classes than in smaller ones. The teacher did notknow I was focusing specifically on corrective feedback; the onlyrequest was that the lessons observed involve the same material andactivities with both groups.

2.4. Data-collection instruments and procedures

The data derives from an observational study of two classes takingthe same EFL course. An interview with the teacher was also held afterclassroom observation was completed.5

2.4.1. Classroom observation

The research was carried out through a structured observation ofteacher/students interaction (real time and recorded) under normalclassroom conditions.6 Three lessons with each class were observedduring the last three weeks of the course with the consent of both theteacher and the students. Real time observation consisted in taking noteswhich focused on errors, verbal feedback and any paralinguistic featuresused to provide feedback such as intonation, stress, gestures, facialexpressions or the teacher writing on the board. When dealing withpair/group work, at least one ‘sub-group’ was observed during eachlesson. The lessons were also video-recorded for subsequent analysis. Thevideo recorder was placed at the front of the classroom and was moved torecord pair and group work when necessary. The recordings consist in 6one-hour lessons altogether, 3 lessons with Group A and 3 with Group B.

2.4.2. Interview

Data regarding the teacher’s views on error treatment were gatheredthrough an interview which took place after classroom observation had

52


5 I had originally intended to give the students a questionnaire on their attitude towards errorcorrection, but was unable to do so due to time constraints.

6 The students did not seem to be self-conscious about the presence of an observer and videocamera as is shown by the fact that some of them carried out part of the activities in Italian ratherthat in English even though they were being observed (see below).

been completed. The interview focused on principles of languageteaching and learning which might influence her treatment of errors, onher thoughts regarding corrective feedback, on how she deals witherrors (e.g. what kind of errors she feels are important to correct, whenand how she corrects these errors). She was also asked to watchexcerpts of her lessons and comment on the way she treated errors.

2.5. Methods of processing

The video recordings were carefully transcribed, using the notestaken during direct observation when necessary.7 Because the first tenminutes of each lesson basically consisted in greeting students as theycame in and answering questions about the course or exam whilewaiting for all students to arrive, they were not transcribed. Therefore,the total length of the recordings transcribed amounts to 5 hours (6‘fifty- minute’ lessons). The categories used to code the data are type ofclassroom activity, type of error, source of error, and type of feedback.

2.5.1. Types of classroom activities

Each lesson was subdivided into the different types of activitiescarried out following the COLT (Communicative Orientation ofLanguage Teaching) category definitions- part A (Fröhlich, Spada andAllen 1985). Activities were, first of all, divided according to theorganization of participation (whole class, group work, pair work,individual). There was often not a clear distinction between pair workand group work since, generally speaking, the students preferredworking in groups and so some of the ‘pair work’ was carried out bymore than two students. Therefore activities involving pair work andgroup work were put into the same category. A similar situationoccurred with work that was supposed to be carried out individually.Once again, students sometimes tended to work together and there wastherefore some interaction. It is for this reason that data regardingactivities that were supposed to be carried out individually but were nothave also been taken into consideration.

The activities were then categorized into activity types through a

53


7 See appendix for transcript conventions used.

description of what participants did (e.g. teacher introduces activity,students prepare for activity, students ask for/give information) andaccording to their content (including the explicit focus on language:form/function/sociolinguistic/discourse and the topics covered:holidays, famous people, personal information). The last categoryregards the materials used, which includes the type of materials (texts(minimal or extended) or visual), the source of the materials: pedagogic(specifically designed for L2 teaching), and the use of materials (highlycontrolled, semi-controlled, minimally controlled).

2.5.2. Types of errors

No distinction was made between ‘errors’ and ‘mistakes’ as it wouldhave been difficult to determine this, with the exception, perhaps, ofcases in which students self-corrected immediately. Therefore the word‘error’ is used in a broad sense to include both terms. Errors weredivided into 5 categories: grammatical, lexical, phonological, contentand unsolicited uses of L1. Examples of each error type are shownbelow.

Grammatical: This includes errors in the use of determiners,prepositions, pronouns (i.e. closed classes); errors in pluralization;errors in tense, in verb morphology, in subject/verb agreement, in theuse of auxiliaries, in question formation and in word order. Forexample:

(1) S1: In a tent. Where did you go?S2: In Greece. (preposition)

(2) S: When John Lennon was born? (question formation)

Lexical: This category includes the inaccurate, imprecise, orinappropriate choice of nouns, verbs, adverbs, and adjectives (i.e. openclasses). For example:

(3) S: Her surname is De Caro. She plays swim.

(4) S1: How did you travel?S2: By flight.

54


Phonological: Coding of utterances regarded phonemes and stress,but did not include rhythm, intonation or pitch range. I concentrated onmispronunciation that can lead to misunderstanding (e.g. ‘ship’ insteadof ‘sheep’, although it is true that the context usually clarifies anymisunderstanding), on pronunciation that was unintelligible and onmispronunciation that could in the long run cause what Porter (1998:A7) refers to as “undue discomfort for the listener”. However, as thislast point may be subjective since, after all, what one person finds‘irritating’ may not be so for others, a native speaker of English wasasked to watch some extracts of the videos and note down anypronunciation errors in order to verify inter-rater reliability. Thecomparison yielded a high level of agreement. Examples ofpronunciation errors include:

(5) T: Yes, playing tennis. And after? What’s the man doing?S: Skiing. /skaing/

(6) S: You traveled /trævəled/ by ship and you stayed on the beach.

Content: This includes utterances in which students provide wrong,incomplete or unnecessary information. For example:

(7) S: She’s from Lamezia Terme. She likes ‘Top of the pops’. Thereare three bedrooms. (incomplete information)T: Where?

Unsolicited uses of L1: This includes instances when students useItalian where they should use English. This category was added to theother four categories after observation of the first lesson, when it wasfound that the students were using much more Italian than expected. Itwas therefore important to include them in the error count. Examples ofunsolicited uses of Italian include:

(8) T: (…) Autumn. But there is another word for autumn. Do youremember?… What happens in autumn?Ss: Cadono le foglie. [= The leaves fall]

(9) S: It wasn’t really hot, the temperature was like, like it is here inwinter. It was ‘mite’. [=warm]

55


Following Lyster (1998), uses of Italian solicited by the teacher werenot considered, for example:

(10) T: Perfect. Ok, we have different categories. ‘Personalinformation’ What’s your name? … your age … ‘Free time’,what’s ‘free time’?Ss: Tempo libero.T: Perfect, ok. Then we have ‘flat or house’. Flat?Ss: Appartamento.

Nor was the use of Italian to give or ask for clarification orexplanations considered as an error. For example:

(11) S1: How old he has? (asks another student) È giusta? [=Is thisright?]S2: Sì, penso di sì. [=Yes, I think so.]S1: (Calls teacher) È giusta ‘how old he has’? [=Is ‘how old hehas’ right?]

Sometimes students used Italian instead of English to carry out thewhole activity. This was not taken into account since there was nointeraction in English.8 For example, in the following case the studentswere supposed to ask each other questions about a past holiday, but didso in Italian:

(12) S1: Dove sei andato? [=Where did you go?]S2: In Spagna. [=To Spain]S1: Con chi? [=With whom?]S2: Con amici. [=With friends]S1: Quando? [=When]S2: Ad agosto. [=In August]

Multiple errors: Finally, when multiple errors occurred in anutterance, each error was coded separately. For example:

56


8 This highlights one of the problems of pair work. Although the teacher was very active andconstantly moved around the classroom, she was not aware that the students had not fulfilled thetask since when they were asked to report back to the class they did so in English.

(13) S: How old he has? (grammatical and lexical)

(14) S: What made John and Paul /pol/ in 1960? (lexical, grammaticaland phonological)

2.5.3. Types of feedback

Feedback was coded following the categories identified in Lysterand Ranta’s (1997) model of corrective feedback. All feedback wastaken into consideration, whether it was provided by the teacher or byanother student. In fact, peers provided feedback in the form of recasts,explicit correction and repetition. Therefore a distinction is madebetween feedback given by the teacher and feedback given by a peer inthese three categories. Moreover, attempts at correction by peers wereincluded as feedback. A category for self-correction was also included.Focus was verbal as well as paralinguistic (i.e. intonation, stress orgestures used in giving feedback).

Explicit correction: The teacher or peer clearly says that what thestudent has said is wrong and provides the correct form. For example:

By teacher:

(15) S: Where is she from? Quindi [=So] … She from…T: Don’t eat the verb. She’s from …

(16) S: Where are you born?T: No, where were you born?

By peer:

(17) S1: How. How old does he?S2: Però al passato. [=But in the past] (Starts discussing the rulesof the past simple in Italian) How old was he? Penso. [=I think]

(18) S1: When did John Lennon was killed?S2: No. When was John Lennon killed.

Attempt at explicit correction by peer:

57


(19) S1: When you born?S2: No, when was you born?S1: No, mi sembra sbagliata. [=No, it seems wrong]S2: Allora [=Then], when was born you?S1: Ok.

Recasts: The student’s utterance is reformulated, but without directlyindicating that it was wrong. For example:

By teacher:

(20) S: Got a fine.T: Yes, he got a fine, got a fine. Ha preso una multa [= he got afine] ... There is a man. Where is he going?Ss: Doctor.T: To the doctor’s.

By peer:

(21) T: Skiing, yes (writes it on the board). The boy is skiing. What isthe next person doing?S1: Windsurf.S2: Windsurfing.

(22) S1: Poi [=Then] …Who did he met in 1966?S2: meet.S1: Sì, [=Yes] meet, meet.

Some recasts were also used as confirmation checks, and this wasspecified in the coding. For example:

(23) S: Irene went to the sea in St. Tropez last year. She went with herbest friend. She went with … by car and stay in a hotel, doexcursions and sunbathe.T: She went on many excursions? (recast + confirmation check)

Clarification requests: The teacher indicates that the student’sutterance has not been understood or that it contains a mistake and thatthe student needs to repeat it. For example:

58


(24) T: The boy is swimming. What is the girl doing?S: Playing to tennis.T: Sorry?

Metalinguistic feedback: The teacher asks questions, givesinformation or makes comments on the formation of the student’sutterance. She may also provide grammatical explanations. Examples:

(25) S: When John Lennon was born?T: (writes it on the board) Do you like this question?

(26) S: What did … What they did in 1960?T: (whispers) Auxiliary, subject, verb.

Elicitation: the teacher directly elicits a reformulation by askingquestions or by pausing to allow the student to complete the teacher’sutterance. For example:

(27) T: Yes, she’s sunbathing. What are the two tourists doing?S1: Visiting.T: There is a word when you visit different places. Do youremember? When you go and visit places...S2: Sightseeing.

(28) S: She went parents.T: She went …S: with her parents.

Repetition: The teacher or peer repeats the student’s error andchanges intonation to draw attention to it. For example:

By teacher:

(29) S: She was born Catanzaro.T: She was born Catanzaro?

By peer:

(30) S1: What’s her name? (not clear who she’s referring to)S2: What’s her name??S1: Di tua sorella. [=Your sister’s]

59


Multiple feedback: This includes a combination of feedback in oneteacher turn. For example:

(31) T: Windsurfing (writes it on the board). And after windsurfing?Ss: Barca a vela. [=Sailboat]T: Yes, barca a vela. In the famous paragraph about Al Wheeler,yes? He went sai….? You don’t remember? ... Sailing …(Elicitation + recast)

(32) T: She’s writing postcards (writes it on the board). And there is anice man, what’s he doing?S: Jogging.T: He’s jogging? He’s not really jogging. What is he doing?(repetition, metalinguistic clues and elicitation)

While agreeing with Lyster and Ranta (1997) that somecombinations such as ‘elicitation + recast’ or ‘recast + metalinguisticfeedback’ could in effect be considered as ‘explicit correction’ ratherthan ‘multiple feedback’, I felt it was important to keep the distinction.As a consequence, all combinations of feedback are included in thiscategory.

Self-correction: Students self-correct without receiving anyfeedback.

(33) S: (…) Later we go… no, we went swimming.

(34) S: Who did he marry /mari/… marry in 1962? (correctspronunciation)

(35) S1: Come si dice si è divertito? [=How do you say ‘he had a goodtime’?]S2: To fun. No, to enjoy himself

No correction: no corrective feedback is given.

2.6. Methods of analysis

The following steps were taken to analyze the data:

60


1. Corrective feedback in each lesson: The data for Group A and GroupB were initially analyzed separately. The first step was to establishthe distribution of error types within each lesson. Each category oferror was therefore expressed as a percentage of total errors made inthe lesson. The rate of corrective feedback given was thenestablished by comparing the number of errors made with thenumber of corrective feedback moves. Finally, in order to establishwhat types of errors the teacher tried to correct and what type offeedback she provided, the number and percentage of feedbackmoves per error type were analyzed.

2. Comparison of corrective feedback in different activity types: Thedata from the different activities were compared in order to establishwhether the teacher corrected different types of errors depending onthe type of activity which was being carried out.

3. Comparison of corrective feedback given to both groups: In order toestablish whether corrective feedback also depends on the student’sproficiency level, a comparison of the same type of activities carriedout with the two groups was made. The rate of corrective feedbackgiven was compared as were the number and percentage of feedbackmoves per error type.

4. Teacher interview: The teacher’s perceptions of how she treatederrors were compared with what she actually did in the classroom.

3. Results and Discussion

3.1. Results

3.1.1. Results Group A

Tables 1-6 present the results of all Group A lessons observed. Table1 shows the distribution of error types in all three lessons. As can beseen, the most frequent error type is grammatical, followed byphonological errors, lexical errors, unsolicited use of Italian and, finally,content errors.

61


Table 1: Group A total: Distribution of error types

The distribution of feedback moves across the different error types isshown in Table 2. The proportion of error types receiving correctivefeedback reflects the distribution rate of the different error types.

Table 2: Group A total: Distribution of feedback moves per error type

Table 3 presents the percentage of errors treated per error type.Overall, corrective feedback was provided for 67.1% of the errorsmade.9 It is important to underline that this figure includes correctivefeedback provided by the teacher, by peers as well as self correction. Ifwe consider each of these categories separately, we find that the teacherprovided feedback for 57.8% of the errors made, peers providedfeedback for 7.6% of the errors while students self corrected 1.7% ofthe errors made.

62


9 The percentages per lesson were as follows: lesson 1 64.2%, lesson 2 68.2%, lesson 3 70.8%.

Error type Number of errors (N=237) Percentage of errors

Grammatical 128 54%

Phonological 58 24.5%

Lexical 26 11%

Unsolicited uses of L1 16 6.7%

Content 9 3.8%

Error Type Feedback moves (N=159) Percentage of FM

Grammatical 90 56.6%


Lexical 22 13.8%


Content 7 4.4%

Table 3: Group A total: Rate of corrective feedback per error type

The distribution of the different feedback types is illustrated in Table4. Overall, 137/159 feedback moves (86.2%) were by the teacher,18/159 (11.3%) by peers, and 4/159 (2.5%) were self correction.

Table 5 presents the distribution of feedback types per error type. Itis interesting to see that peers provided more explicit correction than theteacher (8 moves vs. 7 moves), although it is necessary to point out thatexplicit correction by peers included 3 attempts.

63


Error Type Feedback moves Percentage of errors treated

Unsolicited uses of L1 15/16 93.7%

Lexical 22/26 84.6%

Content 7/9 77.8%

Grammatical 90/128 70.3%

Phonological 25/58 43.1%

Total 159/237 67.1%

• by teacher 137/237 57.8%

• by peer 18/237 7.6%

• self correction 4/237 1.7%

Feedback type Feedback moves (N=159) Percentage of FM

Recast 68 42.8%

• by teacher 60* 37.8%

• by peer 8 5%

Elicitation 32 20.1%

Explicit correction 18 11.4%

• by teacher 9 5.7%

• by peer 9** 5.7%

* one with confirmation check** including attempts

Table 4: Group A total: Distribution of feedback types

The percentage of errors treated varied according to the type ofactivity. The highest percentage of errors was treated during thepresentation of activities to the whole class (91.2%). This was followedby 68.4% of errors treated during pair work which involved preparationof questions and 63.6% of errors treated during a whole class activity inwhich students checked the questions they had prepared. These twofigures show that the percentage of corrective feedback given washigher during pair/group work than during a whole class activity,although both were focusing on accuracy. This can be explained by the

64


Multiple feedback 17 10.7%

• Elicitation + recast 5

• Recast + elicitation 4

• Repetition + elicitation 2

• M. clues + elicitation 2

• Repetition + recast 1

• Repetition + m. clues 1

• Elicitation +explicit correction 1

• Repetition +m. clues + elicitation 1

Clarification request 10 6.3%

Metalinguistic clues 5 3.1%


• by peer 1 0.6%

Repetition 5 3.1%

Self correction 4 2.5%

65


Fee

dbac

k ty

peG

ram

mat

ical

Pho

nolo

gica

lL

exic

alU

nsol

icit

ed u

ses

of L

1C

onte

nt(N

=90)

(N=2

5)(N

=22)

(N=1

5)(N

=7)

Rec

ast

32 (

35.5

%)

21 (

84%

)7*

(31

.8%

)8

(53.

3%)

0

• by

teac

her

29 (

32.2

%)

19 (

76%

)5

(33.

3%)

• by

pee

r3

(3.3

%)

2 (8

%)

3 (2

0%)

Elic

itat

ion

16 (

17.8

%)

1 (4

%)

6 (2

7.3%

)4

(26.

7%)

5 (7

1.4%

)

Ex.

cor

rect

ion

15 (

16.7

%)

02

(9.1

%)

01

(14.

3%)

• by

teac

her

7 (7

.8%

)1

(4.5

5%)

• by

pee

r8*

* (8

.9%

)1*

*(4.

55%

)

Mul

tipl

e fe

edba

ck8

(8.9

%)

06

(27.

3%)

3 (2

0%)

0

Cla

rifi

cati

on r

eque

st7

(7.8

%)

2 (8

%)

00

1 (1

4.3%

)

ML

clu

es5

(5.5

%)

00

00

• by

teac

her

4 (4

.4%

)

• by

pee

r1

(1.1

%)

Rep

etit

ion

5 (5

.5%

)0

00

0

Self

cor

rect

ion

2 (2

.2%

)1

(4%

)1

(4.5

%)

00

*on

e w

ith c

onfi

rmat

ion

chec

k**

incl

udin

g at

tem

pts

Tab

le 5

: Gro

up A

tota

l: D

istr

ibut

ion

of f

eedb

ack

type

s pe

r er

ror

type

low rate of correction of phonological errors during the latter activity(see Table 6). Corrective feedback followed 60.5% of errors madeduring pair/group work in which students asked each other questions,while 50% of the errors made were treated during a reporting activitycarried out with the whole class. Finally, it is not surprising to see thatonly 20% of the errors made when students were working individuallywere treated.

Table 6 illustrates the rate of feedback per error during the differentactivity types. As we can see, during the presentation phase, all errortypes received a high percentage of corrections. In the other types ofactivities, phonological errors tended to have a low rate of correction,whereas grammatical errors were corrected quite frequently (varyingfrom 81.5%- 60.9%), except for those made during the activity carriedout individually. A same trend was found with lexical errors: moreerrors were corrected during accuracy activities and fewer duringfluency activities, although the percentages were higher compared togrammatical errors. Unsolicited uses of L1, instead, always received ahigh rate of correction, regardless of the activity type. Finally, exceptfor the error made when students were asking and giving information,content errors also had a high rate of correction.

The type of feedback given also varied according to the activitytype. As regards feedback given while carrying out activities with thewhole class, of the 62 errors treated during the presentation activities,66.1% were recast, 34/41 by the teacher and 7/41 by a peer, 19.4% werefollowed by elicitation, 6.4% by multiple feedback, 4.8% byclarification requests, 1.6% by both metalinguistic clues and repetitionby the teacher, and none were self corrected. As regards the activity inwhich the students’ questions were checked, 34.3% of the errors werefollowed by recasts, of which 31.4% of the total by the teacher, 20%were followed by elicitation, 17.1% by multiple feedback, 8.6% byexplicit correction and another 8.6% by clarification requests, 5.7% bymetalinguistic clues (equally provided by teacher and peers), and 2.9%each by repetition and by self correction. Errors treated during reportingactivities were followed by recasts in 33.3% of the cases, by multiplefeedback 19% of the time, 14.3% by elicitation, by clarification requestsand by repetition, and 4.8% by metalinguistic clues. There was no selfcorrection.

66


As regards pair/group work, errors treated during the preparation ofquestions were explicitly corrected 69.2% of the time. What isinteresting here is that explicit corrections by peers made up 53.8% ofthe total, although 3/7 were attempts. The remaining errors werefollowed by elicitation (23.1%) and a recast (7.1%). Pair/group workwhich involved asking each other questions showed a different trend.26.9% of the errors treated were followed by recasts, another 26.9%were followed by elicitation, 19.2% were followed by explicitcorrection (15.4% of the total by the teacher and 3.8% through anattempt by a peer), 11.5% were followed by multiple feedback, 7.6% byself correction, 3.8% by clarification requests and another 3.8% bymetalinguistic clues. As regards the activity carried out individually, ofthe two errors treated, one was explicitly corrected by a peer while theother was self corrected.

Table 6: Group A: Types of errors treated per activity type

67


Activity type Grammatical Phonological Lexical L1 Content

Whole class (1) 24/26 14/18 10/10 11/11 3/3

(presentation) 92.3% 77.7% 100% 100% 100%

Pair/ group work (1) 12/18 0 1/1 0 0

(preparing questions) 66.7% 100%

Whole class (2) 22/27 8/22 2/2 0 3/4

(checking questions) 81.5% 36.4% 100% 75%

Pair/ group work (2) 16/26 2/6 5/6 3/4 0/1

(asking for/giving info) 61.5% 33.3% 83.3% 75% 0%

Whole class (3) 14/23 1/12 4/5 1/1 1/1

(reporting) 60.9% 8.3% 80% 100% 100%

Individual 2/8 0 0/2 0 0

25% 0%

3.1.2. Results Group B

Tables 7-12 present the results of all Group B lessons observed.Table 7 shows the distribution of error types in all three lessons.

Table 7: Group B total: Distribution of error types

The distribution of feedback moves across the different error typesis shown in Table 8. Unlike Group A, the proportion of error typesreceiving corrective feedback does not completely reflect thedistribution rate of the different error types. Feedback for phonologicalerrors, in fact, made up a lower percentage than that for unsoliciteduses of L1, although the number of phonological errors was higher (25vs. 8).

Table 8: Distribution of feedback moves per error type

68


Error type Number of errors (N=181) Percentage of errors


Lexical 27 14.9%



Content 3 1.6%

Error Type Feedback moves (N=60) Percentage of FM


Lexical 8 13.3%


Phonological 2 3.3%

Content 2 3.3%

Table 9 presents the rate at which each error type received feedback.The overall low rate of corrective feedback (33.1%)10 was much lowerthan the rate given to Group A (67.1%). This rate was clearly influencedby the low percentage of phonological errors treated (8%), althoughgrammatical errors and lexical errors also received a relatively low rateof feedback. Once again it is important to underline that overall rate ofcorrective feedback includes corrective feedback provided by theteacher, by peers as well as self correction.

The distribution of the different feedback types is illustrated in Table10. As with Group A, recasts were the most popular type of correctivefeedback, making up 43.3% of all feedback moves. Of these, 38.3%were provided by the teacher, while 5% were provided by a peer.Explicit correction made up 16.7% of the feedback moves. This figurewas influenced by the high number of explicit corrections by peers,15% of the total, whereas explicit correction by the teacher made uponly 1.7%. Another interesting figure regards self correction, whichmakes up 15% of the feedback moves.

Table 9: Group B total: Rate of feedback per error type

69


10 The percentages per lesson were as follows: lesson 1 31.8%, lesson 2 40%, lesson 3 31.7%.

Error Type Feedback moves Percentage of errors treated

Content 2/3 66.6%

Unsolicited uses of L1 5/8 62.5%

Grammatical 43/118 36.4%

Lexical 8/27 29.6%

Phonological 2/25 8%

Total 60/181 33.1%

• by teacher 37/181 20.4%

• by peer 14/181 7.7%

• self correction 9/181 5%

* three with confirmation checks** including attempts

Table 10: Group B total: Distribution of feedback types

Table 11 presents the distribution of feedback types per error type.The highest percentage of errors was treated when questions studentshad prepared were checked with the whole class (83.3%). This wasfollowed by 75% of errors treated during group work which involved

70


Feedback type Feedback moves (N=60) Percentage of FM

Recast 26* 43.3%


• by peer 3 5%

Explicit correction 10 16.7%


• by peer 9** 15%

Self correction 9** 15%

Elicitation 5 8.3%


• by peer 1 1.7%

Multiple feedback 5 8.3%

• Repetition +recast by peer 1

• Elicitation + recast 1

• Elicitation + recast by peer 1

• Repetition+ explicit correction 1

• Repetition +recast by peer +

explanation by teacher (m. c.) 1

Repetition 3 5%


• by peer 1 1.7%

Clarification request 2 3.3%

Metalinguistic clues 0 0%

preparation of questions. There were much lower rates for the otheractivity types. 39% of errors were treated when students reported to thewhole class while corrective feedback followed 22.2% of errors madeduring pair/group work in which students asked each other questions.When students worked in pairs/groups and tried to guess each other’sholidays, only 10% of the errors made were treated. Finally, when thewhole class checked the information they had completed after aninformation gap activity, none of the errors made were treated.

* one with confirmation check** including attempts

Table 11: Group B total: Distribution of feedback types per error type

71


Feedback type Grammatical Phonological Lexical Unsolicited Content(N=43) (N=2) (N=8) uses of L1 (N=2)

(N=5)

Recast 13* (30.2%) 2 6 (75%) 5 (100%) 0

• by teacher 10 (23.3%) 2 (100%) 6 (75%) 5

• by peer 3 (6.9%) 0 0 0

Explicit correction 9 (20.9%) 0 0 0 1 (50%)

• by teacher 1 (2.3%) 0

• by peer 8** (18.6%) 1 (50%)

Self correction 7** (16.3%) 0 2** (25%) 0 0

Elicitation 5 (11.6%) 0 0 0 0

• by teacher 4 (9.3%)

• by peer 1 (2.3%)

Multiple feedback 5 (11.6%) 0 0 0 0

Repetition 2 (4.7%) 0 0 0 1 (50%)

• by teacher 2 (4.7%) 0

• by peer 0 1 (50%)

Clarification request 2 (4.7%) 0 0 0 0

ML clues 0 0 0 0 0

• by teacher

• by peer

Table 12 illustrates the rate of feedback per error during the differentactivity types. As we can see, the rate of correction of grammaticalerrors depended on the type of activity. Rates were much higher duringaccuracy activities (75-100%). Phonological errors were never treatedwhen students asked each other questions or when the whole classchecked information with the teacher, and only a low percentage (22%)was treated when students were reporting to the class. The rate ofcorrection of lexical errors was also relatively low, 41.2% whenstudents were reporting back to the class and 25% during pair work.Unsolicited uses of Italian received a high rate (71.4%) when studentswere reporting. Finally, the two content errors made during pair workwere corrected, whereas the error made when checking questions wasnot.

Table 12: Group B: Types of errors treated per activity type

72



Whole class (1) 5/5 0 0 0 0/1

(checking questions) 100% 0%

Group work (1) 3/4 0 0 0 0

(preparing questions) 75%

Whole class (2) 25/67 2/9 7/17 5/7 0

(reporting) 37.3% 22.2% 41.2% 71.4%

Pair/ group work (2) 9/34 0/9 1/8 0/1 2/2

(asking for/giving info) 26.5% 0% 25% 0% 100%

Pair/group work (3) 1/10 0 0 0 0

(guessing a holiday) 10%

Whole class (3) 0 0/6 0/1 0 0

(checking information) 0% 0%

Finally, the type of feedback given varied according to the activitytype. Feedback given to errors made while checking questions with thewhole class included multiple feedback (60%), and explicit correctionby peer (20%) and self correction (20%). Errors treated during thereporting of information to the whole class were followed by recasts59% of the time, by self correction, including two attempts, in 15.4% ofthe cases, by elicitation 10.3% of the time, while multiple feedback,elicitation and metalinguistic clues each accounted for 5.1% of thecorrective feedback given. No corrective feedback was given when thewhole class checked the information they had completed after aninformation gap activity.

As regards pair/group work, feedback given during the preparationof questions was provided equally by recast by peer, explicit correctionby peer and self correction (33.3%), with no teacher intervention. Errorstreated during activities in which students asked each other questionswere followed by explicit correction 66.6% of the time, with explicitcorrection by peers accounting for 58.3% of the total. Recasts by peer,self correction, elicitation by peer, and repetition by peer each accountfor 8.3% of the feedback given. Once again, there was no teacherintervention. Finally, when students were guessing each other’s holiday,the only error that was treated was through a recast by a peer.

3.1.3. Comparison of corrective feedback given to both groups

As was reported in the previous sections, the overall rate ofcorrective feedback provided to Group A (67.1%) was much higherthan in Group B (33.1%). As regards the distribution of the differentfeedback types, recasts were the most common feedback types in bothgroups: 42.8% for Group A (37.8% of the total by the teacher and 5%by peers), and 43.3% for Group B (38.8% of the total by the teacher and5% by peers) (Table 13). However, the distribution of the other types offeedback was different. For example, in Group A, elicitation accountedfor 20.1% of the feedback given, whereas in Group B this figure wasonly 8.3%. Another difference can be noticed as regards self correction.Only 2.5% of the feedback moves in Group A were self correction,whereas in Group B, self correction accounted for 15% of the errortreated.

73


Table 13: Distribution of feedback types per group

However, although these figures are very interesting, I decided toconcentrate on a comparison of the same type of activities carried outwith both groups. This is because, while the topic of each lesson wassimilar for each group, the activities carried out were not always thesame due to the different proficiency levels. It is therefore moresignificant to focus on the same activities in order to avoid differences

74


Feedback type Group A Group B

Recast 42.8% 43.3%

• by teacher 37.8% 38.3%

• by peer 5% 5%

Ex. correction 11.4% 16.7%

• by teacher 5.7% 1.7%

• by peer 5.7% 15%

Elicitation 20.1% 8.3%

• by teacher 20.1% 6.7%

• by peer 0% 1.7%

Multiple feedback 10.7% 8.3%

Self correction 2.5% 15%

Clarification request 6.3% 3.3%

Repetition 3.1% 5%

• by teacher 3.1% 3.3%

• by peer 0% 1.7%

ML clues 3.1% 0%

• by teacher 2.5%

• by peer 0.6%

which could influence the validity of the data. The four activity typescompared were:

1. Pair/group work (i): preparing questions2. Pair/group work (ii): asking each other questions3. Whole class (iii): reporting4. Whole class (iv): checking questions

Table 14 presents a comparison of the data regarding the percentageof errors treated in each activity type. The most notable differencebetween the two groups is the rate of feedback given during pair/groupwork in which students asked each other questions (ii): 60.5% forGroup A and only 19.5% for Group B. This can be partly explained bythe fact that students in Group A called on the teacher more often.Another noticeable difference is the percentage of errors treated whenthe whole class was checking the questions they had prepared (iv):63.6% Group A and 83.3% Group B. The high number of pronunciationerrors which received no corrective feedback in Group A may be one ofthe reasons for this (see Table 15).

Table 14: Percentage of errors treated per activity per group

75


Activity type Group A Group B

Pair/ group work (i)

(preparing questions) 100% (6/6) 75% (3/4)

Pair/ group work (ii)

(asking for/ giving information) 60.5% (26/43) 19.5% (8/41)

Whole class (iii)

(reporting) 50% (21/42) 39% (39/100)

Whole class (iv)

(checking questions) 63.6% (35/55) 83.3% (5/6)

Total errors treated 60.3% (88/146) 55/151 36.4%

Table 15 outlines a comparison between the rate of error correctionper type of error during the different activities. For ease of reference,the data for Group A are written in italics. We can see that, except forthe activity in which the whole class checked questions, the rate of

76




(p. questions) 5/5 0 1/1 0 0

100% 100%


(p. questions) 3/4 0 0 0 0

75%


(ask/give info) 16/26 2/6 5/6 3/4 0/1

61.5% 33.3% 83.3% 75% 0%


(ask/give info) 6/25 0/6 1/8 0/1 1/1

24% 0% 25% 0% 100%

Whole class (iii)

(reporting) 14/23 1/12 4/5 1/1 1/1

60.9% 8.3% 80% 100% 100%

Whole class (iii)

(reporting) 25/67 2/9 7/17 5/7 0

37.3% 22.2% 41.2% 71.4%

Whole class (iv)

(c. questions) 22/27 8/22 2/2 0 3/4

81.5% 36.4% 100% 75%

Whole class (iv)

(c. questions) 5/5 0 0 0 0/1

100% 0%

Table 15: Errors treated per activity per group

correction of grammar errors is higher in Group A. In fact, if weconsider the rate for all the activities, we see a significant difference:70.4% correction rate for Group A and 38.6% for Group B. As regardsphonological errors, a higher percentage of errors were corrected inGroup B when students were reporting to the class. Nevertheless, theoverall rate of correction of phonological errors is higher in Group A:27.5% to 13.5%, although both rates are relatively low. Lexical errorsare also corrected more often during activities carried out with Group Aand their overall of rate of correction is also much higher: 85.7% to32% in Group B. This is also true for L1 errors, although the differenceis not as great: 80% for Group A and 62.5% for Group B. Finally, asregards content errors, overall, the rate is 66.7% for Group A and 50%for Group B. However, since the number of L1 and content errors isvery low, these figures may not be very significant.

Table 16 outlines the type of corrective feedback given to each groupduring the different activity types. Data for Group A are in italics. Whatis noticeable here is that a wider variety of feedback moves is seen inactivities carried out with Group A. For example, when students wereasking for and giving information (ii), all the different types of feedbackwere used, except for repetition. We can also see that there is moreteacher intervention in both types of pair work activities in Group A,due to the fact that students asked for the teacher’s help more often. Asregards activities carried out with the whole class, once again we see amuch wider range of feedback moves when students were checkingtheir questions in Group A, although it is true that the teacher did usemultiple feedback moves with Group B. As regards the reporting phase(iii), Group B has a very high rate of recasts (59%) compared to 33.3%in Group A. Another difference is the rate of self correction, which ismuch higher for Group B. In fact, we only see self correction by GroupA students on two occasions in pair/group work (ii).

77


78


Fee

dbac

k ty

peP

air/

Pai

r/P

air/

Pai

r/W

hole

Who

leW

hole

Who

legr

oup

Gro

upgr

oup

grou

pcl

ass

clas

scl

ass

clas

sw

ork

(i)

wor

k (i

)w

ork

(ii)

wor

k (i

i)(i

ii)

(iii)

(iv)

(iv)

AB

AB

AB

AB

Rec

ast

16.7

%33

.3%

26.9

%12

.5%

33.3

%59

%34

.3%

0

by te

ache

r16

.7%

026

.9%

033

.3%

59%

31.4

%

by p

eer

033

.3%

012

.5%

00

2.9%

Elic

itat

ion

33.3

%0

26.9

%12

.5%

14.3

%10

.3%

20%

0

by te

ache

r33

.3%

26.9

%0

14.3

%10

.3%

20%

by p

eer

00

12.5

%%

00

0

Exp

. Cor

r.50

%33

.3%

19.2

%50

%0

08.

6%20

%

by te

ache

r0

015

.4%

12.5

%8.

6%0

by p

eer

50%

33.3

%3.

8%37

.5%

020

%

Mul

tipl

e fe

edba

ck0

011

.5%

019

%5.

1%17

.1%

60%

Cla

rifi

c. r

eque

st0

03.

8%0

14.3

%5.

1%8.

6%0

ML

clu

es0

03.

8%0

4.8%

05.

7%0

by te

ache

r3.

8%1

2.9%

by p

eer

00

2.9%

Rep

etit

ion

00

012

.5%

14.3

%5.

1%2.

9%0

by te

ache

r0

14.3

%5.

1%2.

9%

by p

eer

12.5

%0

00

Self

cor

rect

ion

033

.3%

7.6%

12.5

%0

15.4

%2.

9%20

%

Tab

le 1

6: C

ompa

riso

n of

type

of

feed

back

per

act

ivity

type

per

gro

up

3.1.4. Teacher interview

One of the main points that the interview highlighted was that theteacher believes corrective feedback is useful and necessary, althoughshe feels that teachers should provide such input in a way that does notinterfere with the flow of conversation. In fact, while watching thevideos, she often pointed out how she tried not to interrupt. Moreover,she said that she used to provide more feedback when she startedteaching, but realized that at times it was disrupting. She also said thathow much she corrects also depends on the type of activity beingcarried out (e.g. focussed on fluency vs accuracy).

Moreover, while watching the videos, she realized that many timesshe had not made herself quite clear and was surprised that studentssometimes seemed to have understood when in fact they had not. Thishighlights the importance of being aware that even if teachers providethe correct form, students may not always pick it up.

The teacher also noticed that she used the board to focus on errorsless than she thought and that she didn’t go over the errors made at theend of the class as much as she thought she did. She feels that peersshould also provide corrective feedback, but worries that they maysometimes interrupt too much. She believes it is important to correctbeginners more than intermediate students in order to help them learnthe correct forms right from the start and not risk confusing both thestudent who makes an error as well as the other students.

3.2. Discussion of results

3.2.1. Group A

The overall rate of corrective feedback provided by the teacher,peers and self correction was 67.1%. The teacher provided feedback for57.8% of all of the errors made. This figure is similar to the findings inLyster and Ranta (1997) and Lyster (1998), 62% and 61% respectively,but slightly higher than the rate in the study by Lyster and Panova(2002), which was 48%. What is interesting is that while peers play arole in the correction of errors (11.3% of the feedback was provided bypeers), self correction accounted for only 2.5% of the moves. In factstudents only self corrected 4/237 errors made (1.7% of total errors): 2

79


grammatical errors, 1 lexical error and 1 phonological error. This couldmean that learners are not yet ready, or confident enough, to try andcorrect their own errors unless they are pointed out.

As regards the percentage of errors treated for each error type,unsolicited uses of L1 were followed by corrective feedback 93.7% ofthe time, while they only made up 6.7% of the total number of errors.However, this is not surprising since this type of error stands out muchmore than the others. Moreover, many of these errors (11/16) weremade during the presentation of the first activity, in which the teacherwas focusing on specific holiday-related vocabulary. On the other hand,only 43.1% of phonological errors received feedback, although theywere the second most common type of error.

Recasts were the most popular type of feedback (42.8%), followed byelicitation (20.1%). Only 1 recast out of 68 was also a confirmationcheck, implying that it is a combination of feedback rarely used by theteacher. This is unfortunate because confirmation checks might helpstudents notice errors more easily. Both recasts and elicitation were alsooften used in multiple feedback moves, confirming that they were thepreferred type of feedback used by the teacher. In fact 10/17 multiplefeedback moves contained recasts, while 14/17 contained elicitation. Thethird type of feedback provided was explicit correction (11.4%). This lastfigure is influenced by the number of explicit corrective moves by peers(5.7% of the total). In fact, when peers correct each other, they seem toprefer explicit correction. This preference might be due to the fact thatsince many students prefer to be explicitly corrected (Kaufmann 1993,cited in James 1998), they might find it useful to provide this type offeedback as well.

Turning to the distribution of feedback types per error type, it isinteresting to see that recasts were the most common for all types oferror except for content errors, which (perhaps predictably) werefollowed mostly by elicitation (71.5%). It should also be highlightedthat all of the different types of feedback were used when correctinggrammatical errors. This might be due to the fact that when one type offeedback did not work, the teacher tried other types. For example:

(36) S: She was born Catanzaro.T: She was born Catanzaro? (repetition)S: Catanzaro.

80


T: Preposition? (metalinguistic clues)S: In Catanzaro.

The findings also show that the percentage of errors treated isinfluenced by the pedagogical focus of the lesson. Not surprisingly,errors made during activities which focused on accuracy received morecorrective feedback. However, the activity in which the questionsstudents had prepared were checked presents a relatively low rate ofcorrection (63.6%) compared to the other activities carried out with thewhole class which also focused on accuracy (91.2%). This lower rate isdue to the fact that the teacher felt it was more important to focus on thegrammar and therefore corrected a low percentage of phonologicalerrors (36.4%). Moreover, this rate (63.6%) was also lower than the rateof correction for errors made during the preparation of questions ingroups (68.4%), although they were both accuracy activities. Thisshows that these students received a relatively high percentage of errorfeedback during group work and that, once again, explicit correctionplayed an important role in the feedback given. This is true even whenthe feedback is only an attempt, because it nevertheless makes studentsthink about the language and prompts a revision (see James 1998). Anexample of this is the following excerpt:

(37) S1: When you born?S2: No, when was you born?S1: No, mi sembra sbagliata. [=No, it seems wrong]S2: Allora [=Then], when was born you?S1: Ok.

S2 later calls the teacher for confirmation:

(38) S2: (Calls teacher) When was you born?T: Remember. I was, you (gestures back), you …S2: were?T: yes.S2: When were born you?T: When were you born?

Although it is the teacher who finally provides the correct form, thestudents are carefully focusing on the language.

81


In the other types of activities (Whole class (1, 2, 3), pair/groupwork (2)), regardless of the pedagogical focus, recasts were the mostpopular form of feedback, followed by elicitation, although thepercentages vary. In the presentation activities, recasts accounted for66.1% of the feedback, although the teacher used a wide variety offeedback types.

Let’s now turn to the question of whether teachers correct differentkinds of errors depending on the type of activity. As expected,grammatical errors had a higher rate of correction during accuracyactivities, especially in those carried out with the whole class (92.3%and 81.5%), although the rate of correction of grammar was not lowduring the fluency activities (61.5% and 60.9%). As regardsphonological errors, these had a low correction rate, regardless ofwhether the activity was carried out in groups or with the whole class orwhether the focus was accuracy or fluency. The only exception is forthe rate of correction of phonological errors made during thepresentation phase, where the correction rate was high (77.7%). Lexicalerrors were corrected 100% of the time during accuracy activities, andbetween 80-83.3% during fluency activities. Both L1 errors and contenterrors had a 100% correction rate in activities carried out with the wholegroup, regardless of the pedagogical focus. L1 errors presented a lowerrate of correction during pair/group focusing on fluency (75%). Finally,the one content error made when students were asking and givinginformation, was not corrected. However, the low number of contenterrors makes it difficult to consider this a general trend.

3.2.2. Group B

The overall rate of error correction for Group B was quite low,which reflects what the teacher said about providing more feedback tobeginners. The teacher, in fact, provides feedback for only 20.4% of theerrors made, with another 12.7% being provided by peers or selfcorrection, for an overall total of 33.1%. It is interesting to note, thatunlike what was seen in Group A, lexical errors received a rather lowpercentage of correction (29.6%). However, it was the low rate ofcorrective feedback given to pronunciation errors (8%) that had thegreatest effect on the overall percentage of errors treated.

82


As with Group A, recasts were the most popular form of correctivefeedback. Three of the 26 recast moves were also confirmation checks,which is a higher percentage than the one in Group A, although it is stillrelatively low. After recasts, the type of corrective feedback most usedwas explicit correction. This is quite significant because, once again,this figure was greatly influenced by peers whose explicit correctionmade up 15% of the total corrective feedback. Overall, both peer andself correction made up a considerable part of the corrective feedbackprovided, 23.3% and 15% respectively. Therefore, compared to GroupA, students are beginning to take a more active role in correctingthemselves and each other, which is what should be aimed for. In one ofthe lessons, peers even provided more feedback than the teacher. In fact,they provided 58.3% of the feedback moves, while teachers provided25%, the remaining 16.7% being self corrections. Overall, students selfcorrected 9 of the 181 errors they made, 7 of which were grammaticalerrors and 2 were lexical. Although these only represent 5% of the totalerrors made, it is higher than the figure in Group A. Two interestingexamples are the following:

(39) S: He traveled by train and he went in Bologna, in summer. It wasvery hot and he had sunbath, sunbathe in Piazza Maggiore.T: He sunbathed.S: He sunbathed in Piazza Maggiore.

In this case the student manages to correct the lexis, although he usesthe wrong tense, which he then corrects after the teacher’s recast.

(40) S1: Where did you go for your last holiday?S2: I go…S1: No, last.S2: I last... I went to Paola.

Although the feedback given by S1 was not clear at first, itnevertheless prompted a correction.

As regards the distribution of feedback types per error type, recastswere once again the most common type of corrective feedback, exceptfor content errors, which were followed by either explicit correction or

83


repetition. As with Group A, a wide variety of feedback types were usedto correct grammatical errors, whereas phonological errors andunsolicited uses of Italian were always followed by recasts, and lexicalerrors were also followed by self correction.

Turning to percentage of errors treated in the different activity types,there is a significant difference between activities which focus onaccuracy, those which focus on fluency, and the activity which focuseson comprehension. As regards both accuracy and fluency, there is also adifference between activities carried out in groups and those carried outwith the whole class, with a higher percentage of errors being correctedduring the ‘whole class’ activities. Finally, during the activity carriedout to check if students had correctly completed the information gap,none of the errors made were treated. The teacher explained that shewas concentrating on whether or not the content was correct andtherefore disregarded the other types of errors. As expected grammaticalerrors were corrected more often during accuracy activities. Of the 24phonological errors made only 2 were corrected. This was whenstudents were reporting to the class (22.2% correction rate), otherwisethe pronunciation errors were ignored. Lexical errors were correctedmore often during the reporting activity (whole class) than whenstudents were asking each other questions, implying that the type ofclassroom organization may affect the rate of feedback of these errors.As regards L1 errors, they received a high rate of correction (71.4%)when students were reporting, although it was not as high as expected.Once again, the number of content errors is too small to be able toestablish a trend.

Finally, the type of feedback also varied according to the activitytype, but unlike Group A, there was no type of feedback that was muchmore popular.

3.2.3. Comparing proficiency levels

As mentioned above, the main difference between the two groups,considering all of the data, is that Group B has a much lower rate ofcorrective feedback and a higher participation rate by students in bothpeer and self correction. However, it is important to verify this trend byanalyzing activities which were the same for both groups.

84


The analysis of the four different types of activities carried out byboth groups shows that the overall rate of correction is indeed muchhigher in Group A (60.3%) than in Group B (36.4%). These figuresconfirm the teacher’s affirmation that she tends to correct beginnersmore often.

In both groups the percentage of errors corrected was affected by thepedagogical focus of the lesson, where more corrective feedback wasgiven during accuracy activities. As regards pair/ group work (i)(preparing questions), a low number of errors was made in both groups(Group A: five grammar errors and one lexical error; Group B: fourgrammar errors), but the percentage of correction was the highest:100% for Group A and 75% for Group B. The reason for the higherpercentage in Group A was that there was some teacher intervention,whereas in Group B there was none. The second highest percentage ofcorrection was in the whole class (iv) (checking questions) activity.However, there was quite a difference between the two groups (GroupA 63.6%, group B 83.3%), due to two main factors: the low rate ofphonological errors corrected in Group A, which can be explainedbecause the teacher was concentrating more on the grammar, and themuch lower number of errors made by Group B students (6 vs. 55)which could have made them more salient to the teacher.

As regards the fluency activities, the differences are quitesignificant, especially in pair/ group work (ii) (asking for/ givinginformation), where the rate of corrective feedback in Group A is threetimes higher than Group B’s (60.5% to 19.5%). There are two possiblereasons for this. First of all, the students in Group B produced muchmore language and as a consequence also more errors. In fact, in GroupA students used Italian to carry out most of the activity. Secondly, therewas also much more teacher intervention in Group A whereas in GroupB only 1/8 feedback moves was by the teacher. Another interestingdifference is that a much wider range of feedback was provided inGroup A, in fact the only technique that was not used was repetition.Finally, in whole class (iii) (reporting), we again see a higher rate ofcorrective feedback in Group A (50% to 39%). This can be explained inpart by the fact that students in Group B produced longer sentenceswhich the teacher was hesitant to interrupt and, at the same time,wanted to reinforce correct lexis used.

85


4. Conclusion

This case study has highlighted the wide variety of feedbacktechniques used to push students to produce the target form, althoughrecasts are the most common. Furthermore, it has shown that there is asignificant difference between the rate of feedback provided to the twogroups. Moreover, although self correction is low in both groups, it ishigher in the pre-intermediate group, implying that these learners arebeginning to play a more active role in correcting their errors. Finally,the study has looked at the role played by the students themselves inproviding corrective feedback to each other, which could be the topicfor further research. In that case, in addition to focusing on peercorrection, it would be interesting to investigate the difference betweenlearner uptake and type of feedback as well as to compare uptake in thedifferent groups.

86


References

Allwright, Richard L., 1975, “Problems in the Study of the Language Teacher’sTreatment of Learner Error”. In: Burt, Marina K. / Dulay, Heidi C. (eds.), OnTESOL ’75: New Directions in Second Language Learning, Teaching andBilingual Education, Washington D.C., TESOL, 96-109.

Allwright, Dick / Bailey, Kathleen, 2004, Focus on the Language Classroom,Cambridge, CUP.

Bartram, Mark / Walton, Richard, 2001, Correction. A Positive Approach toLanguage Mistakes, London, Language Teaching Publications.

Brown, Douglas H., 1987, Principles of Language Learning and Teaching,Englewood Cliffs NJ, Prentice-Hall.

Chaudron, Craig, 1977, “A Descriptive Model of Discourse in the CorrectiveTreatment of Learner’s Errors”. Language Learning 27/1: 29-46.

Chaudron, Craig, 1998, Second Language Classrooms, Cambridge, CUP.

Doughty, Catherine / Varela, Elizabeth, 1998, “Communicative Focus on Form”.In: Doughty, Catherine / Williams, Jessica (eds.), Focus on Form in ClassroomSecond Language Acquisition, New York, CUP: 114-138.

Fanselow, John, 1977, “The Treatment of Error in Oral Work”. Foreign LanguageAnnals 10/4: 583-593.

Fröhlich, Maria / Spada, Nina / Allen, Patrick, 1985, “Differences in theCommunicative Orientation of L2 Classrooms”. TESOL Quarterly 19/1: 27-56.

Han, Zhaohong, 2002, “A Study of the Impact of Recasts on Tense Consistency inL2 Output”. TESOL Quarterly 36/4: 543-572.

Hendrickson, James, 1978, “Error Correction in Foreign Language Teaching:Recent Theory, Research, and Practice”. In: Croft, Kenneth (ed.), Readings onEnglish as a Second Language. Second Edition. (1980) CambridgeMassachusetts, Winthrop Publishers Inc.: 153-173.

James, Carl, 1998, Errors in Language Learning and Use. Exploring ErrorAnalysis, London, Longman.

Klein, Wolfgang, 1986, Second Language Acquisition, Cambridge, CUP.

Long, Michael, 1977, “Teacher Feedback on Learner Error: Mapping Cognitions”.In: Brown, Douglas H. / Yorio, Carlos A. / Crymes, Ruth H. (eds.), On TESOL’77, Washington D.C., TESOL: 278-293.

Lyster, Roy, 1998, “Negotiation of Form, Recasts, and Explicit Correction inRelation to Error Types and Learner Repair in Immersion Classrooms”.Language Learning 48/2: 183-218.

87


Lyster, Roy / Ranta, Leila, 1997, “Corrective Feedback and Learner Uptake”.Studies in Second Language Acquisition 19/1: 37-66.

Mackey, Alison / Gass, Susan / McDonough, Kim, 2000, “How Do LearnersPerceive Interactional Feedback?” Studies in Second Language Acquisition22/4: 471-497.

Mackey, Alison / Philp, Jenefer, 1998, “Recasts, Interaction and InterlanguageDevelopment: Are Responses Red Herrings?” Modern Language Journal 82/3:338-356.

Oliver, Rhonda, 2000, “Age Difference in Negotiation and Feedback in Classroomand Pairwork”. Language Learning 50/1: 119-151.

Panova, Iliana / Lyster, Roy, 2002, “Patterns of Corrective Feedback and Uptake inan Adult ESL Classroom”. TESOL Quarterly 36/4: 573-595.

Porter, Don, 1998, MATEFL Pedagogic Phonetics and Phonology, Reading,University of Reading.

Tedick, Diane / de Gortari, Barbara, 1998, “Research on Error Correction andImplications for Classroom Teaching”. The ACIE Newsletter 1/3, at<www.carla.umn.edu/immersion/acie/vol1/May1998.pdf> [accessed 13 June2005].

Tonkyn, Alan, 1998, MATEFL Second Language Learning Principles Materials,Reading, University of Reading.

Van Lier, Leo, 1988, The Classroom and the Language Learner, London,Longman.

Vigil, Neddy / Oller, John, 1976, “Rule Fossilization: A Tentative Model”.Language Learning 26/2: 281-295.

88


APPENDIX: TRANSCRIPTION SYSTEM

T: teacherS: student (S1, S2 used to signal that two different students are

speaking)Ss: several students speaking at once… pauses of a few seconds( ) used for writer’s comments[=] used for translations in English/ / phonemic transcription, used only when words are pronounced

incorrectly.XXX incomprehensible“ ” anything read rather than spoken____ word is given extra stress- used to indicate that there is a longer pause between each word

89


Corrective Feedback in EFL university classrooms: A Case Study at an Italian … · 2017-12-14 ·...

Documents

Transcript of Corrective Feedback in EFL university classrooms: A Case Study at an Italian … · 2017-12-14 ·...